crypto: implement ARM AES backend for Shavite512's CompressElement()

kwvg · kwvg · commit 963215bf9172 · 2025-09-22T09:52:06.000Z
diff --git a/src/Makefile.am b/src/Makefile.am
@@ -786,7 +786,8 @@ crypto_libbitcoin_crypto_arm_aes_la_CXXFLAGS += $(ARM_AES_CXXFLAGS)
 crypto_libbitcoin_crypto_arm_aes_la_CPPFLAGS += -DENABLE_ARM_AES
 crypto_libbitcoin_crypto_arm_aes_la_SOURCES = \
   crypto/x11/arm_crypto/aes.cpp \
-  crypto/x11/arm_crypto/echo.cpp
+  crypto/x11/arm_crypto/echo.cpp \
+  crypto/x11/arm_crypto/shavite.cpp
 
 # See explanation for -static in crypto_libbitcoin_crypto_base_la's LDFLAGS and
 # CXXFLAGS above
diff --git a/src/crypto/x11/arm_crypto/shavite.cpp b/src/crypto/x11/arm_crypto/shavite.cpp
@@ -0,0 +1,38 @@
+// Copyright (c) 2025 The Dash Core developers
+// Distributed under the MIT software license, see the accompanying
+// file COPYING or http://www.opensource.org/licenses/mit-license.php.
+
+#if defined(ENABLE_ARM_AES)
+#include <crypto/x11/util/util.hpp>
+
+#include <cstdint>
+
+#include <arm_neon.h>
+
+namespace sapphire {
+namespace arm_crypto_shavite {
+void CompressElement(uint32_t& l0, uint32_t& l1, uint32_t& l2, uint32_t& l3,
+                     uint32_t r0, uint32_t r1, uint32_t r2, uint32_t r3, const uint32_t* rk)
+{
+    // Pack block + XOR with round key 1
+    uint8x16_t block = util::pack_le(r0, r1, r2, r3);
+    block = util::Xor(block, vreinterpretq_u8_u32(vld1q_u32(&rk[0])));
+    // AES round + XOR with round key 2
+    block = util::Xor(util::aes_round_nk(block), vreinterpretq_u8_u32(vld1q_u32(&rk[4])));
+    // AES round + XOR with round key 3
+    block = util::Xor(util::aes_round_nk(block), vreinterpretq_u8_u32(vld1q_u32(&rk[8])));
+    // AES Round + XOR with round key 4
+    block = util::Xor(util::aes_round_nk(block), vreinterpretq_u8_u32(vld1q_u32(&rk[12])));
+    // AES round
+    block = util::aes_round_nk(block);
+    // Unpack + XOR with l values
+    uint32x4_t result = vreinterpretq_u32_u8(block);
+    l0 ^= vgetq_lane_u32(result, 0);
+    l1 ^= vgetq_lane_u32(result, 1);
+    l2 ^= vgetq_lane_u32(result, 2);
+    l3 ^= vgetq_lane_u32(result, 3);
+}
+} // namespace arm_crypto_shavite
+} // namespace sapphire
+
+#endif // ENABLE_ARM_AES
diff --git a/src/crypto/x11/dispatch.cpp b/src/crypto/x11/dispatch.cpp
@@ -49,6 +49,10 @@ void RoundKeyless(uint32_t x0, uint32_t x1, uint32_t x2, uint32_t x3,
 namespace arm_crypto_echo {
 void FullStateRound(uint64_t W[16][2], uint32_t& k0, uint32_t& k1, uint32_t& k2, uint32_t& k3);
 } // namespace arm_crypto_echo
+namespace arm_crypto_shavite {
+void CompressElement(uint32_t& l0, uint32_t& l1, uint32_t& l2, uint32_t& l3,
+                     uint32_t r0, uint32_t r1, uint32_t r2, uint32_t r3, const uint32_t* rk);
+} // namespace arm_crypto_shavite
 #endif // ENABLE_ARM_AES
 
 #if defined(ENABLE_SSSE3)
@@ -165,6 +169,7 @@ void SapphireAutoDetect()
         aes_round = sapphire::arm_crypto_aes::Round;
         aes_round_nk = sapphire::arm_crypto_aes::RoundKeyless;
         echo_round = sapphire::arm_crypto_echo::FullStateRound;
+        shavite_c512e = sapphire::arm_crypto_shavite::CompressElement;
     }
 #endif // ENABLE_ARM_AES
 #endif // !DISABLE_OPTIMIZED_SHA256
